Strengthening Opportunities Along the Meat Supply Chain to Promote Sustainable Agriculture in Intermountain States

RO1. Evaluate how a large, multi-sectoral event such as a meat conference (incorporating at least 400 ranchers/producers, meat processors, and meat supply chain stakeholders) results in measurable changes in connections and relationships throughout the meat supply chain, including across rural and urban stakeholders. 

As the number of opportunities for values-based procurement and claims-based markets grow, for livestock producers to effectively benefit from these opportunities there must be enhanced communication and understanding across supply chains - from production to consumers. This becomes more of a challenge as meat supply chains are highly consolidated (e.g., MacDonald et al. 2000), limiting direct communication between producers and consumers across the supply chain. Carolan (2020), for example, conducted interviews with individuals engaged in shaping urban food policy and institutional procurement plans in Denver, and rural Colorado farmers and ranchers. He found important differences across the two groups in terms of conceptions of “good food” and what it means to be a “good farmer”; reconciling these requires mediation. 

Accordingly, we are proposing the rigorous evaluation of a meat conference, to be held twice during the grant period. The purpose of this research objective is to evaluate how an event that intentionally brings together stakeholders along the meat supply chain and incorporates activities designed to build bridges can result in measurable changes in connections and relationships, including across rural and urban stakeholders. To do this we will follow methods used by Brasier and Goetz (2010) and Love et al. (2020). Brasier and Goetz (2010) analyzed the number of new connections made at a local foods conference, particularly for those with few professional linkages prior to the conference. Love et al. (2020) analyzed how different educational curriculum in a construction management program impacted social networks. 

Data: Data collection will center around the meat conferences held in year one quarter four in Colorado, and year three quarter four in Montana. For each meat conference, registrants will be surveyed before and after the event. We expect that at least 200 individuals will participate in each year of the meat conference, and participation in the pre survey will be required for all attendees to attend the event. To improve the response rate following the event, we will raffle an iPad. Accordingly, we estimate that at least 100 individuals will participate in both the pre and post survey, thus providing useable data for our social network analysis. 

Participants will be asked to identify the types of relationships they have with different types of stakeholders across the meat supply chain before and after the event. For each question, they will add an individual name and organization name with whom they have had a written or verbal exchange about something pertaining to meat within the past year. Who they identify will be piped through to future sections of this survey where they will be asked to answer additional questions about these individuals/organizations. 

Additionally, respondents will be asked questions regarding their view of different claims-based language and labels including “sustainable” “regenerative” and “grass-fed”. This will help us to discern if changes in values, production practices, or market opportunities were related to changing supply chain relationships. Finally, producers will be asked pre/post questions related to pre/post markets, prices, and profitability to enable us to understand if new relationships were formed that impacted sales.

Methods: Social network data will be analyzed statistically using UCInet (Borgatti and Freeman 2002), and social network diagrams will be created using Visone (Brandes and Wagner 2004). Density, a measure of the number of actual connections in a network divided by the number of possible connections (Giuffre 2013), will be used to measure ‘within’ network connections and to explain network cohesion. In this research, we will use density to compare changes from the pre/post surveys. Additionally, we will calculate fragmentation in the network, a whole network statistic that measures the proportion of pairs of nodes that cannot reach each other (Hanneman and Riddle 2005). 

Three network statistics will be used to evaluate the network as well as changes within and across the network. First, average degree is a measure of the number of connections for each actor in the network (Giuffre 2013). Second (described previously), we will calculate the fragmentation in the network. Third, Freeman’s centralization will be used to characterize the overall structure of a network through “in” and “out” measures of centrality. When a network is more centralized, it indicates that the few central individuals are more influential than those on the periphery. In contrast, a network with low centrality shows that social relationships and influence are more evenly balanced across the network. 

Baran (1964) described the difference between three ideal types of network structures: a) a centralized star-like network, b) a decentralized network, and c) a distributed un-centralized network. The key characteristic of a centralized network is its star-like pattern. Centralized star-like networks are vulnerable because the removal of a key central node can cause all other nodes to become disconnected from each other (this is what we might expect to see in the ‘pre’ meat supply chain, as it is generally only the processor or distributor that communicates with both the livestock producer and buyer). A decentralized network also has a star-like pattern, and is still considered a vulnerable or imbalanced network because the central nodes have more influence. In contrast, in a distributed network, all nodes can connect and communicate with other nodes, thus reducing the imbalance of power found in centralized networks. If the meat conference is successful in strengthening ties throughout the supply chain, this is the type of result we would expect to see.

RO2. Examine multiple years of USDA ARMS data and conduct empirical analysis to investigate how participation in local food market channels (i.e., farmers market, other direct, retail, distributor/institution), as well as operation / producer characteristics, and locational variables impact the profitability of livestock producers/ranchers in the Western U.S. 

Recent national research on the profitability impacts of sales through local food markets (e.g., Bauman et al. 2018a, 2018b; Jablonski et al. 2020) provides important preliminary information for producers making determinations about market channel selection. Recent research by Jablonski and Bauman, for example, use USDA ARMS data to create financial benchmarks for farms and ranches with sales through local food markets, provide preliminary estimates of profitability implications of local food marketing strategies, investigate the role of human capital measured in terms of wage rates and labor expenditures, and evaluate the financial efficiency of local food producers (fact sheets based off of peer-reviewed research can be found here: https://localfoodeconomics.com/benchmarks/#factsheetsection). As an example of some of the analysis used in this research, Bauman et al. (2018) divide the sample by gross cash farm income, as well as quartile within each sales class (i.e., quartile four is the most profitable and quartile one the least profitable) to investigate the relationship between key financial characteristics and profitability for farms selling through local markets. Across sales class ($1,000-$74,000, $75,000-$349,999, $350,000-$1M, and >$1M), each of the quartile groups are significantly different at the one percent level. They find that among the top performing quartiles, direct-to-consumer only marketers had lower return on assets (their measure of profitability) compared to top performers using intermediated only markets or a combination of direct-to-consumer and intermediated markets. Bauman et al. (2019) explored the financial efficiency of farms and ranches that sell through local food markets to better understand what factors could improve their viability and performance. Profit efficiency frontier maps the combinations of fixed and variable inputs that are utilized by the most profitable local food producers. They find that, on average, a farm or ranch could increase profit by about 133% by improving efficiency (i.e,. utilizing a different combination of fixed and variable inputs). Overall, most farms and ranches using local food markets are not producing on the efficiency frontier and could realize significant improvements in profitability with changes in their operation. 

Though this preliminary analysis is important - and points to the importance of understanding the relationship between market channel selection and production characteristics - more in-depth analysis is needed to provide information to producers that can meaningfully inform market channel selection. This research objective will utilize the USDA ARMS data, focused on livestock producers in the Western US, to examine the relationship between market channel selection and production, demographic and locational characteristics. 

Data: The USDA ARMS is USDA’s primary source of information on the financial condition, production practices, and resource use of America’s farm businesses and the economic well-being of America’s farm households. The ARMS is a nationally representative survey that targets about 30,000 farms annually, including operations of all sizes and commodities (Kachova 2013; USDA ERS 2020). ARMS data are restricted-access, and Jablonski and Bauman have been approved by the USDA National Agricultural Statistics Service (NASS) and Economic Research Service (ERS) to utilize the data. Access is available via NORC at the University of Chicago’s secured network, which can be accessed remotely.  

Since 2008, the ARMS includes questions about farm sales through local food channels and provides a sufficiently large sample of producers participating in these markets (Low and Vogel 2011). The 2013-2016, and 2019 Phase III data include more detailed questions about local food channels. ARMS participants were asked to report if they produced, raised, or grew commodities for human consumption that were sold directly to (1) individual consumers, (2) retail outlets, or (3) institutions. Subsequently, they are asked to provide the monetary amounts received from selling (1) directly to consumers at farmers markets; (2) directly to consumers from on-farm stores, u-picks, roadside stands, or CSAs; (3) to local retail outlets such as restaurants or grocery stores; (4) to regional distributors such as food hubs; or (5) to local institutional outlets such as schools or hospitals (USDA ERS 2019). However, due to sample size limitations, we will aggregate sales to (4) and (5). 

PI Jablonski and Bauman have utilized these data extensively to analyze profitability impacts of farms and ranches selling through local food markets (Bauman et al. 2018, 2019; Jablonski 2020). However, they have not yet analyzed farm-level profitability implications for ranchers specifically. In addition, the ARMS has more detailed information on cattle compared to other species, so we will leverage these data to provide detailed estimated costs of production. 

Methods: We use two methodological approaches to analyze our data. In both methods, we focus on a subsample of livestock producers/ranchers from ARMS that are located in the Western region of the U.S.. First, we follow Bauman et al. (2019) and divide our sample into quantiles segmented by their profitability performance (defined as return on assets) to understand how strong and weak performance may vary based on market channel decisions, operation / producer characteristics (including variable expenses, labor share of variable expenses, land tenure, beginning farmer status, debt utilization, producer age, producer education, etc.), and locational variables (including ruralness as defined by the rural-urban continuum code). This analysis is largely descriptive, but allows for results that are easily digestible by Extension and producer audiences. 

Second, we use a more rigorous approach following Park (2015): unconditional quantile regression (UQR). UQR allows one to evaluate the impact of changes in the distribution of the explanatory variables (in our case operation / producer characteristics, market channel selection, and location variables) on quantiles of the unconditional (marginal) distribution of an outcome variable (in our case, profitability, defined as return on assets) (Firpo et al. 2009). UQR is an appropriate approach as it enables us to provide more refined estimates of the relationship between the operation/producer characteristics, market channel selection, and location variables, across the entire distribution of return on assets. Though these results will be a bit more complicated to explain to Extension and producer audiences, they will provide important checks on the descriptive findings/analysis.  

For both RO1 and RO2, results will be presented and vetted at academic conferences and published in peer-reviewed journals. Additionally, for RO1 we will present results to extension and other technical assistance audiences with recommendations for future convenings or attempts to strengthen relationships across diverse stakeholder groups. For RO2 we will make results available to extension and producer audiences via factsheets on the localfoodeconomics.com website as well as through the meat school classes and online pricing tool (see education plan below).